ChatGPT for Real Estate Agents: 10 Prompts That Work (and Where They Fail)
ChatGPT for real estate agents is now an everyday tool: in NAR's 2025 Technology Survey, ChatGPT was the AI tool agents used most, named by 58% of those using AI. It can draft a decent listing in seconds, if you give it the right prompt and check what comes back. Here are 10 ChatGPT prompts for real estate agents that work, the mistakes to watch for, and when a dedicated listing tool saves you the effort.
Why most listing prompts fail
"Write a listing for a 3-bed house" produces the same flowery text every agent has seen: "nestled", "boasting", "stunning". The model has nothing specific to work with, so it fills the gaps with clichés, and sometimes with invented details. OpenAI's own research explains why: standard training rewards a confident guess over admitting uncertainty, so language models produce plausible statements that aren't true (OpenAI, 2025). In a property advert, an invented "sea view" or "renovated kitchen" is a real problem.
Three things fix most of it: give the facts, set the rules, and say exactly what you want back.
Prompt 1: the listing description
You are an experienced real estate copywriter in [country/city].
Write a listing description for [portal or MLS] using ONLY the facts below.
Never add features, views, condition claims, distances or neighbourhood
claims that are not listed. If a detail is missing, leave it out.
Facts:
- Type: [semi-detached house]
- Price: [£425,000]
- Bedrooms / bathrooms: [3 / 2]
- Size: [105 m²]
- Features: [south-facing garden, kitchen refitted 2024, driveway]
- Location facts: [7-minute walk to station]
Rules: maximum [1,000] characters including spaces. Lead with the
strongest feature. Describe the property, never the ideal buyer.
No clichés such as "nestled", "boasting" or "stunning".
End with a call to book a viewing.
Why it works: the facts are fenced off, the length is fixed, and the instruction against inventing details is explicit.
Prompts 2 to 5: headlines, social posts, emails and rewrites
Headlines
Using only the facts above, write 5 listing headlines under 70 characters.
Each must name a specific feature. No exclamation marks or ALL CAPS.
Social media posts
Turn this listing into: (1) a Facebook post of 60–100 words with one
call to action, (2) an Instagram caption with short lines and at most
5 relevant local hashtags (Instagram's limit), (3) a 30-second Reels script with on-screen
text. Keep every fact identical to the listing.
Client email
Write a short email to my buyer list announcing this listing. Friendly,
under 120 words, three bullet-point highlights, and ask them to reply
to book a viewing. Use [First name] as the greeting placeholder.
Rewriting your old listing
Here is my current listing. Rewrite it to put the strongest feature
first, cut clichés and keep it under [800] characters. Do not add any
facts that are not in the original. Then list anything you removed.
For ready-made examples of the output you're aiming for, see our real estate listing description examples and just listed post examples.
Prompts 6 to 10: ChatGPT prompts for the rest of your week
The same rule applies outside listings: paste in the facts, tell it not to add any, and say what you want back.
Monthly market update
Write a 150-word market update email for homeowners in [area].
Use ONLY these figures: [homes sold last month, median price,
average days on market, change from last year]. Explain what they
mean for someone thinking of selling, in plain words. No predictions.
Price reduction announcement
This listing's price has dropped from [£450,000] to [£425,000].
Write a short social post and a 2-line email subject and opening that
present it as new value for buyers. Don't call it a bargain or say
the seller is desperate. Keep every other fact as in the listing.
Open house follow-up
Write a follow-up email to people who came to the open house at
[address] on [date]. Thank them, repeat the 3 strongest features from
the listing, mention [any offers deadline], and ask one question that
invites a reply. Under 100 words. Placeholder: [First name].
Reply to a buyer enquiry
A buyer asked: "[paste their message]". Draft a friendly reply that
answers only from the listing facts below, says plainly which
questions I need to check with the seller, and offers two viewing
times: [times]. Facts: [paste listing].
Valuation appointment prep
I'm meeting a homeowner at [address] to value their home. From my
notes below, list 8 questions to ask about the property and their
plans, and 3 points to explain about how I'd market it.
Notes: [what you know so far].
Where ChatGPT still fails
- Invented details. Even with strict prompts, check every claim against your notes. "Close to schools" or "quiet street" can creep in.
- Fair housing wording. Models still write "perfect for families" or "ideal for young professionals". Rules in the US and many other countries treat that as stating a preference (NAR). Ask the model to describe the property only, and read for it.
- Local conventions. A generic prompt won't know that Dutch listings add "k.k." to the price, that UK lettings quote rent per calendar month, or that Irish adverts must state the BER. Spell these out.
- Length. Models count words loosely and characters badly. Check against your portal's limit before you paste.
- Privacy. On consumer ChatGPT plans, conversations can be used to train models unless you opt out (OpenAI Help Center). Don't paste vendors' names, phone numbers or personal circumstances.
- Prompting time. Typing facts into a chat, fixing the tone and asking again adds up across a week of listings.
ChatGPT vs a dedicated listing tool
| Task | ChatGPT | Dedicated listing generator |
|---|---|---|
| Collecting the facts | You type them into a prompt | A form asks for the right details |
| Stopping invented details | Depends on your prompt and checking | Built into the instructions, with a fact check that flags unsupported claims |
| Character limits | Approximate | Set an exact count |
| Posts, email, translations | One prompt each | Written together from one set of details |
| Flexibility for other tasks | Excellent | Listings only |
Use ChatGPT for the open-ended work: market updates, scripts, answering a tricky email. For listings you write every week, our AI listing description generator collects the facts with a form, writes only from them and checks the result. It turns one set of details into the description, headline, highlights, Facebook and Instagram posts, a Reels script, WhatsApp messages and a client email.
If you're comparing options, our guide to the best AI tools for real estate agents covers the wider field. If you're new to all of this, start with AI for real estate agents who aren't tech people.
References
- National Association of Realtors (2025). Realtors embrace AI, digital tools to enhance client service, NAR survey finds.
- OpenAI (2025). Why language models hallucinate.
- OpenAI Help Center. How your data is used to improve model performance.
- National Association of Realtors. Window to the Law: advertising within the fair housing framework.
- Lilach Bullock. ChatGPT prompts for estate agents: property descriptions.